News for hackers. Stuff that matters.
Re:Building a deeper understanding of images by tlarkworthy on Saturday September 06, 2014 @09:18AM (#8278280)
The neural architectures change so a set of parameters for one network won't run on another. Image size changes too which changes the nn behind it. There is not much work on mapping one set of parameters onto another. Transfer learning might have a little applicability. But unlikely at the needing edge of vision research.

Worth noting a neural network IS a general purpose function encoding.

There are no comments on this comment yet.